The goals of this project are two fold. First, the planned experiments will continue our investigation of the role of motor cortex in learning and performing skilled voluntary reaching movements. The second, closely related goal is to identify the extended information coding capabilities of neuronal populations engaged in these behaviors. The experiments will specifically examine how single neurons and neuron populations in monkey motor cortex acquire task specific feature combinations during motor learning. The experiments examine three frontal motor areas, which contain neurons with different feature coding capabilities: the supplementary motor cortex (SMA), associated with learned movement sequences; the premotor area (PMA), associated with learned sensorimotor associations; and the primary motor cortex (MI), associated with movement direction. To examine how task features are acquired and combined into components of a motor action we will record simultaneously from multiple single neurons in MI, PMA, SMA or combinations of these areas, as monkeys perform a well learned action and then acquire new actions. The continuation of two experiments that require learning sensory and motor features combinations or the joining of motor components into sequences is planned. For Exp. 1, sensorimotor adaptation, monkeys learn a new arbitrary mapping between a familiar sensory cue and a known movement. For Exp. 2, motor sequence learning, monkeys learn to combine learned component arm movements into sequences. Longitudinal studies of multiple, simultaneously recorded neurons in behaving monkeys is made possible by new chronically implantable electrode arrays developed within this project. The analysis will characterize the stability of existing feature codes in single cell firing rate and populations, the ability for cells and populations to acquire new feature combinations and the extended coding provided by population interactions. Added information available from population interactions will be evaluated using methods that incorporate temporal or covariance information available only in the simultaneously recorded cell group. Identification of single cell and population contributions to learned motor actions is an important step towards revealing general cortical encoding principles and should also help to establish how the cortex participates in learning new behaviors. The results obtained may also reveal a means to obtain real time control signals from neural population that can be used in assisting neurologically impaired individuals.